<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 12pt;
font-family:Calibri
}
--></style></head>
<body class='hmmessage'><div dir='ltr'>Thanks a lot Matt, I will have a look at those sources.<br>Pablo<br><br><div>> Date: Tue, 5 Jan 2016 22:46:37 -0500<br>> Subject: Re: [ITK-dev] Integrating header-only boost::Compute for faster itkGPU development.<br>> From: matt.mccormick@kitware.com<br>> To: pablo.hernandez.cerdan@outlook.com<br>> CC: insight-developers@itk.org<br>> <br>> Hi Pablo,<br>> <br>> To make it easier to develop ITK-GPU filters with boost::Compute, an<br>> ITK Module should be created. See the Modules in<br>> ITK/Modules/ThirdParty/ as examples. Documentation on how to create a<br>> Module can now be found in the ITK Software Guide:<br>> <br>> http://itk.org/ITKSoftwareGuide/html/Book1/ITKSoftwareGuide-Book1ch9.html#x50-1430009<br>> <br>> If an Insight Journal article is submitted, the module can be merged<br>> into ITK as a Remote Module, and it can be dependency for modules that<br>> want to use boost::Compute in ITK.<br>> <br>> <br>> Other similar libraries that have been discussed include Cuda:<br>> <br>> http://www.insight-journal.org/browse/publication/802<br>> https://code.google.com/p/cuda-insight-toolkit/<br>> https://github.com/SimonRit/RTK/tree/1defcbdfe0bdcac6d0786f62ed8530e1b39204f9/code<br>> <br>> and ArrayFire:<br>> <br>> http://arrayfire.com/<br>> <br>> And similar Modules could be created for these. Each has its own<br>> advantages in terms of platform support and programability. It is<br>> recommended to reuse as much of the existing ITK OpenCL GPU<br>> infrastructure as possible or to use a similar design.<br>> <br>> <br>> To drive ITK-GPU computing forward in general, we need more software<br>> quality dashboard submissions. Instructions on how to set up a Nightly<br>> dashboard submission can be found here:<br>> <br>> http://www.itk.org/Wiki/ITK/Git/Dashboard<br>> <br>> Thanks,<br>> Matt<br>> <br>> <br>> On Sat, Jan 2, 2016 at 9:04 AM, Pablo Hernández<br>> <pablo.hernandez.cerdan@outlook.com> wrote:<br>> > https://github.com/boostorg/compute<br>> > It is like cuda::Thrust library, functional oriented, but using opencl and<br>> > header-only. It implements quite a lot of STL algorithms to run in GPU's. It<br>> > requires c++11, and will be officially available in next boost release<br>> > (1.61)<br>> > I mention this, because I was creating a home-brewed itkGPU filters, and it<br>> > is painful to hand-write opencl code for every basic operation out there. I<br>> > can of course use it by my own, but I thought it could be helpful to boost::<br>> > the GPU developing in ITK.<br>> > I have no conflict of interests with this particular library, but wanted to<br>> > share it and hear what you think. Also any comment about the road map<br>> > involving GPU in ITK would be interesting.<br>> ><br>> > Regards,<br>> > Pablo<br>> ><br>> ><br>> ><br>> > _______________________________________________<br>> > Powered by www.kitware.com<br>> ><br>> > Visit other Kitware open-source projects at<br>> > http://www.kitware.com/opensource/opensource.html<br>> ><br>> > Kitware offers ITK Training Courses, for more information visit:<br>> > http://kitware.com/products/protraining.php<br>> ><br>> > Please keep messages on-topic and check the ITK FAQ at:<br>> > http://www.itk.org/Wiki/ITK_FAQ<br>> ><br>> > Follow this link to subscribe/unsubscribe:<br>> > http://public.kitware.com/mailman/listinfo/insight-developers<br>> ><br></div> </div></body>
</html>